BayesAge 2.0: a maximum likelihood algorithm to predict transcriptomic age.

Area of Science:

• Genomics
• Biomarker Discovery
• Aging Research

Background:

• Aging is a multifaceted biological process influenced by genetic and environmental factors.
• Transcriptomic age (tAge) prediction from RNA-seq data is crucial for understanding aging.
• Existing methods face challenges in accuracy, age bias, and computational efficiency.

Purpose of the Study:

• To introduce BayesAge 2.0, an upgraded maximum likelihood algorithm for predicting tAge.
• To enhance the prediction of biological age using gene expression data.
• To provide a more robust, accurate, and efficient tool for aging research.

Main Methods:

• BayesAge 2.0 integrates a Poisson distribution for count-based gene expression data.
• LOWESS smoothing is employed to capture nonlinear gene-age relationships.
• The algorithm builds upon the original BayesAge framework for epigenetic age prediction.

Main Results:

• BayesAge 2.0 demonstrates significant improvements over traditional linear models like Elastic Net regression.
• Minimal age-associated bias was observed in prediction residuals.
• Reference construction and cross-validation are computationally more efficient than Elastic Net regression.

Conclusions:

• BayesAge 2.0 offers a robust and accurate method for tAge prediction.
• The algorithm addresses key limitations of existing models, including age bias and computational time.
• BayesAge 2.0 is a valuable tool for advancing aging research and biomarker development.